Removal of sequences in pre.cluster

I’ve perform the pre.cluster step on my data and noticed that in the output, it says:
Total number of sequences before precluster was 6297.
pre.cluster removed 2600 sequences.

Does the pre.cluster command MERGE sequences that contains what it thinks are mismatches due to sequencing errors, or does it REMOVE these sequences from the analysis? I do notice that the total number of sequences reads for my samples at the end did decrease, compared to the number of sequences I had after all the filtering steps but prior to pre.cluster. Thanks for any clarification.

It merges them, it removes them from the “unique” sequence pool, but they;re still included as duplicate sequences.